From: Nicolas Pitre Subject: Re: [PATCH v2 0/3] ARM: NEON based fast(er) AES in CBC/CTR/XTS modes Date: Fri, 4 Oct 2013 14:39:29 -0400 (EDT) Message-ID: References: <1380837566-18242-1-git-send-email-ard.biesheuvel@linaro.org> <20131004174853.GY24303@mudshark.cambridge.arm.com> <20131004182353.GQ12758@n2100.arm.linux.org.uk> Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Cc: Ard Biesheuvel , Will Deacon , "linux-arm-kernel@lists.infradead.org" , "linux-crypto@vger.kernel.org" , "patches@linaro.org" To: Russell King - ARM Linux Return-path: Received: from mail-qe0-f42.google.com ([209.85.128.42]:55157 "EHLO mail-qe0-f42.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751043Ab3JDSjc (ORCPT ); Fri, 4 Oct 2013 14:39:32 -0400 Received: by mail-qe0-f42.google.com with SMTP id nc12so481264qeb.15 for ; Fri, 04 Oct 2013 11:39:31 -0700 (PDT) In-Reply-To: <20131004182353.GQ12758@n2100.arm.linux.org.uk> Sender: linux-crypto-owner@vger.kernel.org List-ID: On Fri, 4 Oct 2013, Russell King - ARM Linux wrote: > Also, remember that the GPL says: > > "The source code for a work means the preferred form of the work for > making modifications to it." > > So here's the question: is the assembly code the perferred form to make > modifications? From what you're saying above, the answer to that seems > to be no. Now, I'm not going to throw toys out of the pram and say > that this is a hard requirement, but just take a moment to think about > how we treat vendors who don't do this, instead supplying "non-preferred" > forms of source code, and think about whether there's double standards > here. > > Now, while I can imagine that people have an ideological objection to > perl (using comments like "write only code" etc) that's not a good > enough excuse to avoid including the perferred form for future > modification. > > Now, we have mechanisms in the kernel build where we can include a > prepared source which can be used to lessen the burden on the toolset > required to build the kernel. So, including both the perl script and > the pre-generated assembly is entirely acceptable. This tends to > nullify the excuse that "we don't want to add additional tool burden > to kbuild" argument. Those are very good arguments. And I agree with your recommendation as well. Nicolas